How's GLM4 basic compared to Dirac? If anything it is the other way around.
I want to go into more detail, but I'm done with that. I tried twice to write a big write up comparing it to dirac a few weeks ago in another thread to respond to
@Sancus, but both times my post was lost. FWIW, I think he uses both Dirac(for the bass) and GLM4 for correcting his 8351b.
Many of my issues are usability/UX related, could be personal preference and/or user error related. And, given that it's still in beta, I don't care so much about those:
- Lots of crashing(bricked my 8351B twice). Will likely be fixed before launch.
- Store to monitor function bugged(tried on one windows machine and two macos machines). Will likely be fixed before launch.
- Lack of separation between measurement and correction processes. With Dirac you do your 17 measurements once, and then you never have to measure again(just load the measurements and test different curve files that you feed it).
- No way to redo or stop and save the measurement process. This was really frustrating, as I'd get 8 measurements in and then my dog would start barking, or my neighbors dog would start barking, or a loud truck would drive by, and I'd have to start over from the beginning. After 15+ hours of this, it led to me mostly taking just 3-5 measurements and quitting out of fear after that. I suppose that could partly explain why the results were so poor(confirmed by ear and REW). That said, I did have a few projects that were 15+ measurements, and the results weren't all that much better.
- Default target curve is flat. I suppose this is ideal for studio use?
- No support for custom target curve files. You can adjust the individual filters manually, and the manual target does allow you to customize a bit, but it's no where near as customizable as Dirac. With Dirac you can see each measurement, and feed it a target curve file(which you can further customize with the GUI) to get the target that is best suited for each speaker.
- Only supports Genelec speakers/subs. Not a problem for the speakers, as they're SOTA, and after hearing the 8351b, I'm going all in with Genelec. Does kinda suck for the subwoofers though, as Genelec subs aren't very good for the money if you don't need all those connections. My RS2s(despite being cheaper) sound so much better than my 7370s. GLM only working with the latter is a real pain point.
The more concerning part, though, were the problems with the results:
- Lack of results consistency. Same target, same mic position(didn't touch it), drastically different results, confirmed with REW. The results were
wildly different, to the point I don't even trust what it's telling me.
- Predicted response often looks nothing like actual resulting response(measured with REW). Dirac lies a lot here too, but it's closer, at least according to REW. I was able to drastically increase the measured and heard sound quality by applying additional filters post GLM via REW and miniDSP.
- No sophisticated handling of dips. Dirac seems to use the 17 measurements to determine the cause of the dips. It fixes the ones it can fix, partially fixes the ones it can partially fix, and ignores the ones that it can't do anything about. GLM4 seems to just ignore all the dips, no matter the cause, and then just crush the peaks.
- The sound. This was the main thing, and I could live with all the other stuff if GLM sounded better than Dirac. But, it took me just a couple hours of measuring and trying different targets with Dirac to beat the best results I was able to get with GLM after 30-40 hours of fiddling. 2 whole weekends(+ several days) of messing with GLM, and within a couple hours of starting Dirac, I already had something that sounded and measured(at least according to REW) better. I did also trying using both GLM and Dirac, and while it did sound better than GLM on its own, I couldn't get it to sound as good as Dirac on its own.
All that said, there is definitely something to be said for my lack of experience with GLM. I've had several years now to learn the quirks of Dirac and how to get better results. This is my first experience with GLM ever. Perhaps with equivalent GLM experience, I might get better results. Who knows. I was super enthusiastic about getting to use it, but not so much any more
.
At some point I may give it another go, but probably not for awhile. I'm actually looking at replacing Dirac with a software based Audiolense solution now. From what I've been told by people I trust, it's even better than Dirac. I just installed it on my new machine today and started messing around
.
Something I did like about GLM was the automatic phase alignment when I was still trying to use the 7370s. Dirac 3.0 does that now too, at least with 2 subs, but it does a poor job of pointing it out(which is something I like better with GLM).
If you've got a lot more experience with GLM, I'd be interested in learning. I still have the box and cables lying around
.